1: \begin{abstract}

2: This paper studies the consensus problem of heterogeneous multi-agent systems by the feedforward control and  linear quadratic (LQ) optimal control theory.

3: Different from the existing consensus control algorithms,

4:  which require to design an additional distributed observer for estimating the leader's information

5: and to solve a set of regulator equations.

6: In this paper, by designing a distributed feedforward controller, a non-standard neighbor error system is  transformed into a standard linear system,

7:  and then an optimal consensus controller is designed by minimizing a combined state error with neighbour agents.

8: The proposed optimal controller is obtained by solving Riccati equations, and

9: it is shown that the corresponding cost function under the proposed distributed controllers is asymptotically optimal.

10: The proposed consensus algorithm can be directly applicable to solve the consensus problem of homogeneous systems.

11: %traditional consensus algorithm for homogeneous systems can  be viewed as a special case of the current algorithm.

12: Simulation example indicates the effectiveness of the proposed scheme and a much faster convergence speed than the existing algorithm.

13: %, and a much faster convergence speed than the conventional consensus methods.

14:   %Moreover,  the new method avoids computing nonzero eigenvalues of the communication topology as in the traditional consensus methods.

15: \end{abstract}